Postgresql中xlog生成和清理逻辑操作

时间:2022-10-03 16:42:02

0 前言

1、2部分是对XLOG生成和清理逻辑的分析,XLOG暴涨的处理直接看第3部分。

1 WAL归档

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
# 在自动的WAL检查点之间的日志文件段的最大数量
checkpoint_segments =
# 在自动WAL检查点之间的最长时间
checkpoint_timeout =
# 缓解io压力
checkpoint_completion_target =
# 日志文件段的保存最小数量,为了备库保留更多段
wal_keep_segments =
# 已完成的WAL段通过archive_command发送到归档存储
archive_mode =
# 强制timeout切换到新的wal段文件
archive_timeout =
max_wal_size =
min_wal_size =

1.1 不开启归档时

文件数量受下面几个参数控制,通常不超过

?
1
(2 + checkpoint_completion_target) * checkpoint_segments + 1

checkpoint_segments + wal_keep_segments + 1个文件。

如果一个旧段文件不再需要了会重命名然后继续覆盖使用,如果由于短期的日志输出高峰导致了超过

3 * checkpoint_segments + 1个文件,直接删除文件。

1.2 开启归档时

文件数量:删除归档成功的段文件

抽象来看一个运行的PG生成一个无限长的WAL日志序列。每段16M,这些段文件的名字是数值命名的,反映在WAL序列中的位置。在不用WAL归档的时候,系统通常只是创建几个段文件然后循环使用,方法是把不再使用的段文件重命名为更高的段编号。

当且仅当归档命令成功时,归档命令返回零。 在得到一个零值结果之后,PostgreSQL将假设该WAL段文件已经成功归档,稍后将删除段文件。一个非零值告诉PostgreSQL该文件没有被归档,会周期性的重试直到成功。

2 PG源码分析

2.1 删除逻辑

触发删除动作

?
1
2
3
RemoveOldXlogFiles
> CreateCheckPoint
> CreateRestartPoint

wal_keep_segments判断(调用这个函数修改_logSegNo,然后再传入RemoveOldXlogFiles)

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
static void
KeepLogSeg(XLogRecPtr recptr, XLogSegNo *logSegNo)
{
 XLogSegNo segno;
 XLogRecPtr keep;
 XLByteToSeg(recptr, segno);
 keep = XLogGetReplicationSlotMinimumLSN();
 /* compute limit for wal_keep_segments first */
 if (wal_keep_segments > 0)
 {
 /* avoid underflow, don't go below 1 */
 if (segno <= wal_keep_segments)
  segno = 1;
 else
  segno = segno - wal_keep_segments;
 }
 /* then check whether slots limit removal further */
 if (max_replication_slots > 0 && keep != InvalidXLogRecPtr)
 {
 XLogSegNo slotSegNo;
 XLByteToSeg(keep, slotSegNo);
 if (slotSegNo <= 0)
  segno = 1;
 else if (slotSegNo < segno)
  segno = slotSegNo;
 }
 /* don't delete WAL segments newer than the calculated segment */
 if (segno < *logSegNo)
 *logSegNo = segno;
}

删除逻辑

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
static void
RemoveOldXlogFiles(XLogSegNo segno, XLogRecPtr endptr)
{
  ...
  ...
 while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL)
 {
 /* Ignore files that are not XLOG segments */
 if (strlen(xlde->d_name) != 24 ||
  strspn(xlde->d_name, "0123456789ABCDEF") != 24)
  continue;
 /*
  * We ignore the timeline part of the XLOG segment identifiers in
  * deciding whether a segment is still needed. This ensures that we
  * won't prematurely remove a segment from a parent timeline. We could
  * probably be a little more proactive about removing segments of
  * non-parent timelines, but that would be a whole lot more
  * complicated.
  *
  * We use the alphanumeric sorting property of the filenames to decide
  * which ones are earlier than the lastoff segment.
  */
 if (strcmp(xlde->d_name + 8, lastoff + 8) <= 0)
 {
  if (XLogArchiveCheckDone(xlde->d_name))
        # 归档关闭返回真
        # 存在done文件返回真
        # 存在.ready返回假
        # recheck存在done文件返回真
        # 重建.ready文件返回假
  {
  /* Update the last removed location in shared memory first */
  UpdateLastRemovedPtr(xlde->d_name);
        
        # 回收 或者 直接删除,清理.done和.ready文件
  RemoveXlogFile(xlde->d_name, endptr);
  }
 }
 }
  ...
  ...
}

2.2 归档逻辑

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
static void
pgarch_ArchiverCopyLoop(void)
{
 char xlog[MAX_XFN_CHARS + 1];
  
  # 拿到最老那个没有被归档的xlog文件名
 while (pgarch_readyXlog(xlog))
 {
 int  failures = 0;
 for (;;)
 {
  /*
  * Do not initiate any more archive commands after receiving
  * SIGTERM, nor after the postmaster has died unexpectedly. The
  * first condition is to try to keep from having init SIGKILL the
  * command, and the second is to avoid conflicts with another
  * archiver spawned by a newer postmaster.
  */
  if (got_SIGTERM || !PostmasterIsAlive())
  return;
  /*
  * Check for config update. This is so that we'll adopt a new
  * setting for archive_command as soon as possible, even if there
  * is a backlog of files to be archived.
  */
  if (got_SIGHUP)
  {
  got_SIGHUP = false;
  ProcessConfigFile(PGC_SIGHUP);
  }
  # archive_command没设的话不再执行
      # 我们的command没有设置,走的是这个分支
  if (!XLogArchiveCommandSet())
  {
  /*
   * Change WARNING to DEBUG1, since we will left archive_command empty to
   * let external tools to manage archive
   */
  ereport(DEBUG1,
   (errmsg("archive_mode enabled, yet archive_command is not set")));
  return;
  }
      # 执行归档命令!
  if (pgarch_archiveXlog(xlog))
  {
  # 成功了,把.ready改名为.done
  pgarch_archiveDone(xlog);
  /*
   * Tell the collector about the WAL file that we successfully
   * archived
   */
  pgstat_send_archiver(xlog, false);
  break;  /* out of inner retry loop */
  }
  else
  {
  /*
   * Tell the collector about the WAL file that we failed to
   * archive
   */
  pgstat_send_archiver(xlog, true);
  if (++failures >= NUM_ARCHIVE_RETRIES)
  {
   ereport(WARNING,
    (errmsg("archiving transaction log file \"%s\" failed too many times, will try again later",
     xlog)));
   return; /* give up archiving for now */
  }
  pg_usleep(1000000L); /* wait a bit before retrying */
  }
 }
 }
}

2.3 ready生成逻辑

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
static void
XLogWrite(XLogwrtRqst WriteRqst, bool flexible)
{
...
      if (finishing_seg)
  {
  issue_xlog_fsync(openLogFile, openLogSegNo);
  /* signal that we need to wakeup walsenders later */
  WalSndWakeupRequest();
  LogwrtResult.Flush = LogwrtResult.Write; /* end of page */
        # 归档打开 && wal_level >= archive
  if (XLogArchivingActive())
          # 生成ready文件
   XLogArchiveNotifySeg(openLogSegNo);
  XLogCtl->lastSegSwitchTime = (pg_time_t) time(NULL);
...

2.4 总结

ready文件只要满足archive_mode=on和wal_lever>=archive,就总会生成(XLogWrite函数调用生成)

因为archive_command设置空,所以ready文件的消费完全由外部程序控制

done文件的处理由PG完成,两个地方会触发done文件处理,检查点和重启点

处理多少done文件受wal_keep_segments和replication_slot控制(KeepLogSeg函数)

3 WAL段累积的原因(长求总?)

注意:无论如何注意不要手动删除xlog文件

注意:checkpoint产生的日志回不立即生成ready文件,是在下一个xlog后一块生成的

3.1 ReplicationSlot

打开流了复制槽

?
1
2
3
4
5
-- 流复制插槽
-- 如果restart_lsn和当前XLOG相差非常大的字节数, 需要排查slot的订阅者是否能正常接收XLOG,
-- 或者订阅者是否正常. 长时间不将slot的数据取走, pg_xlog目录可能会撑爆
select pg_xlog_location_diff(pg_current_xlog_location(),restart_lsn), *
from pg_replication_slots;

删除

?
1
select pg_drop_replication_slot('xxx');

删除后PG会在下一个checkpoint清理xlog

3.2 较大的wal_keep_segments

检查参数配置,注意打开这个参数会使xlog和ready有一定延迟

3.3 回收出现问题

如果不使用PG自动回收机制,数据库依赖外部程序修改.ready文件,需要检测回收进程

?
1
(archive_mode=on archive_command=''

3.4 检查点间隔过长

检查参数配置

以上为个人经验,希望能给大家一个参考,也希望大家多多支持服务器之家。如有错误或未考虑完全的地方,望不吝赐教。

原文链接:https://blog.csdn.net/jackgo73/article/details/90108958